Detecting action units on 3D faces

Automatic facial action unit (AU) detection is a research topic that finds many applications in behavioral science and human computer interaction. The AU detection performance in 2D images are maturing but are not yet adequate. In this study, we develop a method to detect AUs in 3D images and show its superiority vis-a-vis 2D. The data modality is 2D curvature map, which is obtained by conformal mapping of 3D surface data. Since performance comparisons are run on 2D data with the same algorithms, any bias that could be induced by 3D modality is precluded. We address also the choice of generative versus discriminative classifiers, and consider 2D-3D fusion.